A more predictable capital calculation is useful. Mistaking it for permission to stop asking hard risk questions is not.
On September 30, the Federal Reserve finalized two changes to its supervisory stress-test framework: greater transparency and public input, plus averaging to reduce year-to-year volatility in stress capital buffers. The Fed separately proposed a revision to its noninterest-income model; that item is not a final rule.
For affected banks, the practical task is to connect the new framework to model governance, data lineage and capital planning. A smoother number does not eliminate the need to explain the decisions behind it.
What is final, and when it matters
The first final rule requires annual public input on stress-test scenarios and material model changes, updates scenario-design arrangements and adopts models for the 2027 test. For banks with large trading books, the Fed will use two global market shock components, with each firm's results reflecting the component that produces the larger losses for that firm.
The second final rule averages the two most recent annual supervisory stress-test results when calculating stress capital buffers for firms tested in both years. The averaging begins in 2028, so the calculation uses models that incorporate public input.
These are not interchangeable dates. Treat the 2027 models and 2028 averaging as distinct change-management items with their own dependencies.
Less volatility is not a capital holiday
The Fed expects the changes to reduce year-over-year volatility in capital requirements by approximately 50%, without materially changing aggregate capital requirements. That is an expected framework-level effect, not a promise of a 50% reduction in a particular bank's required capital.
The distinction belongs in the board paper, not in a correction after the meeting. No institution should treat that projection as evidence that its own underwriting capacity, distributions, or risk appetite can change immediately.
For planning teams, compare the implications under the institution's own scenarios and constraints. Ask which movements come from portfolio changes, which from models, and which from the calculation method.
If those effects cannot be separated, an apparently calmer result could conceal a weaker understanding of the business. Predictability should improve planning, not lower the standard of explanation.
Transparency changes the evidence burden
Annual consultation creates a reason to organize technical feedback as an operating process. Decide who reviews scenarios, who assesses material model changes, and who can substantiate the institution's concerns.
Make each material observation traceable to an assumption, a data source, and an accountable reviewer. “The model is conservative” is not a particularly useful submission if nobody can explain where, why, or under which conditions.
This is where policy becomes a practical advantage. A repeatable process for model approval, challenge, and exceptions is more useful than scrambling to assemble one after a consultation opens.
The governance lesson also travels beyond directly affected banks. Smaller institutions can borrow the discipline of traceable assumptions without claiming that these specific rules apply to them.
Build on the risk infrastructure you already have
Begin with existing finance, risk, and data systems. Test whether the same portfolio can be reconciled across reporting views and whether model versions, scenario assumptions and manual adjustments are recorded consistently.
Where those foundations work, extend them rather than creating another detached analysis environment. Where they do not, fix the ownership and lineage problems before layering on more automation.
A controlled digital twin of the planning workflow can help test changes to versions and exception paths. That means reproducing the logic and controls with appropriately protected or synthetic data, not pooling confidential portfolios in pursuit of a more impressive demonstration.
The test should be operational: can the team reproduce a result, explain the differences and identify who approved the change? A fast answer that cannot be reconstructed is not a governance improvement.
Keep the separate proposal separate
The non-interest-income proposal would better capture differences in how banks generate fee income. Its comment window runs for 60 days after Federal Register publication, not automatically from the September 30 press-release date.
Maintain distinct workstreams for adopted requirements, open proposals, and the bank's own choices. That prevents an unfinalized model change from quietly becoming a production assumption.
The Fed has changed the framework. Bank leaders still have to make the operating model explainable.
FAQs
Did the Fed finalize all three changes?
No. The September 30 announcement finalizes two rules, covering transparency and stress-capital-buffer averaging. The revision of the non-interest-income model remains a separate proposal.
When does averaging begin?
The two-year averaging starts in 2028 for firms subject to the supervisory test in both relevant years. The release also identifies models adopted for the 2027 test, which is a separate milestone.
Does the expected 50% reduction mean banks need half as much capital?
No. The Fed's estimate concerns year-over-year volatility in capital requirements, not a halving of capital levels. The Fed does not expect a material change in aggregate requirements.
What should CIOs and risk leaders do first?
Create a shared change inventory covering models, scenarios, data, approvals and reporting. Keep final requirements and proposed changes separately labelled, and require reproducible results before altering production workflows.